Geiser, Henchoz, Galland, Garrupt, Veuthey
2374 Laurent Geiser1 Yveline Henchoz1, 2 Alexandra Galland2 Pierre-Alain Carrupt2 Jean-Luc Veuthey1 1
Laboratory of Pharmaceutical Analytical Chemistry, School of Pharmaceutical Sciences, EPGL, University of Geneva, Geneva, Switzerland 2 LCT – Pharmacochemistry, School of Pharmaceutical Sciences, EPGL, University of Geneva, Geneva, Switzerland
Determination of pKa values by capillary zone electrophoresis with a dynamic coating procedure CZE allows to measure the acidic dissociation constant (pK a) of many drug substances. However, determining the EOF intensity may be time-consuming, especially at a low pH. In order to overcome this drawback, a dynamic coating procedure of the capillary was carried out to increase lEOF, and thus to reduce the analysis time. In addition, this coating procedure enhanced migration time stability. The effective mobilities of 15 compounds were measured at different pH, producing pK 9a values dependent on BGE ionic strength. The latter values were corrected with the activity coefficient to obtain a “true” pK a value. The 15 investigated compounds were (i) five acids: namely, salicylic acid, benzoic acid, ketoprofen, phenobarbital, and phenol, (ii) four bases: lidocaine, propafenone, propranolol, and quinine, (iii), five ampholytes: sulfanilamide, sulfabenzamide, sulfadimethoxine, sulfadoxine, and sulfisoxazole, and (iv) one zwitterion: cetirizine. The range of determined pK a values was between 1.2 and 11.2, and close to the pK a values available from the literature. Key Words: Capillary zone electrophoresis; Dissociation constants; Dynamic coating; Sulfonamides; Received: May 12, 2005; revised: June 14, 2005; accepted: June 16, 2005 DOI 10.1002/jssc.200500213
1 Introduction In order to determine the pharmacokinetic and pharmacodynamic profiles of a compound, some physicochemical properties are needed, such as solubility, lipophilicity, molecular size, hydrogen-bonding capacity, and charge [1]. The latter is easily calculated at any pH once acidic dissociation constants (pK a) are determined [2]. The thermodynamic acid dissociation constant (K a) of a monoprotic acid, HA, is defined as Ka ¼
cHþ cA cHA
½Hþ ½A ½HA
log cA ¼
pffiffi 0:51 z 2 I pffiffi 1þ I
pKa ¼ pH log
½A log cA ½HA
ð2Þ
Activity coefficients for ions in a dilute electrolyte solution at 258C can be estimated from the Debye – Hckel theory [4]. This model requires the knowledge of the hydrated ion diameter, which is generally an unknown value comprised Correspondence: Professor Jean-Luc Veuthey, Laboratory of Pharmaceutical Analytical Chemistry, School of Pharmaceutical Sciences, EPGL, University of Geneva, Bd d’Yvoy 20, 1211 Geneva 4, Switzerland. Fax: +41-22-379-6808. E-mail:
[email protected].
J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
ð3Þ
where I is the solution ionic strength (M), and z the ion charge. By combining Eqs. (2) and (3), the following equation is obtained for monoprotic acids
ð1Þ
where c terms are activity coefficients and terms in brackets represent molar concentrations. Given that activity coefficient of HA is assumed as unity [3], and the pH of the solution as – log (cH+ [H+]), Eq. (1) can be related to the pK a pKa ¼ pH log
between 1 and 10 . As a substitute, the Gntelberg model [5] can be used, which estimates this unknown value as equal to 3 . So, the activity coefficient can be written as
pffiffi ½A 0:51 z 2 I pffiffi þ ½HA 1þ I
ð4Þ
A similar equation is also valid for monobases pKa ¼ pH log
pffiffi ½B 0:51 z 2 I p ffiffi ½BHþ 1þ I
ð5Þ
Traditionally, potentiometric and spectrophotometric titration are the methods of choice to determine pK a values [2]. Since 1990, CZE has been considered as an interesting alternative to measure pK a, such as demonstrated in numerous articles (e. g., [3, 6 – 10]). In fact, this method exhibits attractive features, recently pointed out by Poole et al. [2]. The method is simple and can be easily automated; sample and solvent consumption are negligible; samples may contain some impurities; compounds of low water solubility can be handled, and no exact concentration knowledge is required.
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Determination of pKa values by CZE During a CZE analysis, the acidic nature of the silica material provokes a flow in the fused-silica capillary, which initiates a solution motion called EOF moving from the anode toward the cathode. As a result, the analyte apparent mobility (lapp; cm2/V6s) is the sum of the effective mobility (leff) and the mobility of EOF (lEOF). The latter is strongly dependent on the BGE composition, especially its pH. In fact, elevated lEOF occurs at high pH (pH A5.0), while lower lEOF are measured at low pH (pH a4.0). Consequently, determining leff requires measuring lEOF, generally with a neutral marker (e. g., acetone). Practically, leff of an analyte can be calculated as
1 1 tm tmn
analyses), which can decrease the precision on t m and t mn determination. Moreover, Joule heating is often important with short capillaries and high voltage, as it increases the temperature inside the capillary. Short-end injection may also afford a drawback related to temperature; because the capillary end size is not thermostated, results are less reliable. In this paper, 15 compounds, shown in Fig. 1 (five acids, four bases, five ampholytes, and one zwitterion), were investigated by CZE to determine pK a values. To prevent the aforementioned drawbacks, a dynamic coating procedure was performed for a strong and constant EOF whatever the BGE pH.
ð6Þ
where t m and t mn are, respectively, the migration times (s) of the analyte and the neutral marker, V is the applied voltage (V), L tot the total capillary length (cm), and L eff the effective capillary length (cm), i. e., from the injection point toward the detector. The run time per analysis is closely related to the investigated compound charge, and is especially time-consuming if EOF does not rapidly transfer the anionic compounds to the detector. In this particular case, polarity mode and EOF have both been switched [11], performing the detection at the anode outlet and reversing EOF with surfactants. However, this approach is not generic, since it is neither adapted for bases nor for ampholytes. In order to reduce the analysis time for pK a determination, other approaches compatible with all compounds have been used, such as short-end injection [12], use of shorter capillary or pressure-assisted analysis [13 – 15]. Nevertheless, these approaches present some drawbacks. First, they reduce peak efficiency (especially with pressure-assisted
2 Materials and methods 2.1 Chemicals Benzoic acid, salicylic acid, sulfadimethoxine, quinidine, and quinine were purchased from Fluka (Buchs, Switzerland). Ketoprofen, lidocaine, sulfanilamide, sulfabenzamide, sulfisoxazole, propafenone, and propranolol were from Sigma-Aldrich (Buchs, Switzerland). Phenol was obtained from Siegfried (Zofingen, Switzerland), phenobarbital from Haenseler (Herisau, Switzerland), sulfadoxine from Roche (Basel, Switzerland), and cetirizine from UCB (Braine-l’Alleud, Belgium). Analytical reagent grade phosphoric acid (H3PO4), formic acid (HCOOH), acetic acid (CH3COOH), boric acid (H3BO3), CAPS, sodium hydroxide, and acetone were obtained from Fluka. HPLC grade methanol was by Romil (Klliken, Switzerland). The Ceofixm pH 2.5 solutions were purchased from Analis (Namur, Belgium). Ultra-pure water was supplied by a Milli-Q RG purification unit from Millipore (Bedford, MA, USA).
Figure 1. Structures of the 15 investigated molecules. J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Original Paper
leff
Ltot Leff ¼ V
2375
2376
Geiser, Henchoz, Galland, Garrupt, Veuthey
Table 1. BGE of 20 mM ionic strength, according to Phoebus software Buffer composition
Theoretical pH
Measured pH
Buffer capacity, mM/pH unity
Generated current, lA
H3PO4 43.1 mM; NaOH 8.5 mM
2.0
1.97
50.9
44 – 46
H3PO4 27.3 mM; NaOH 16.3 mM
2.5
2.47
20.6
24 – 25
H3PO4 22.3 mM; NaOH 18.8 mM
3.0
2.95
7.4
17 – 18
HCOOH 51.0 mM; NaOH 19.6 mM
3.5
3.46
30.0
18 – 19
HCOOH 29.8 mM; NaOH 19.9 mM
4.0
3.96
15.7
17 – 19
CH3COOH 51.6 mM; NaOH 20.0 mM
4.5
4.48
29.5
16 – 17
CH3COOH 30.0 mM; NaOH 20.0 mM
5.0
4.96
15.6
16 – 17
CH3COOH 23.2 mM; NaOH 20.0 mM
5.5
5.41
6.3
16 – 17
H3PO4 17.0 mM; NaOH 18.5 mM
6.0
5.97
3.3
13 – 14
H3PO4 13.6 mM; NaOH 16.8 mM
6.5
6.44
5.9
13 – 14
H3PO4 10.1 mM; NaOH 15.0 mM
7.0
6.94
6.1
13 – 14
H3PO4 8.0 mM; NaOH 14.0 mM
7.5
7.47
3.5
12 – 13
H3PO4 7.1 mM; NaOH 13.5 mM
8.0
8.05
1.4
12 – 13
H3BO3 114.0 mM; NaOH 20.0 mM
8.5
8.45
40.4
14 – 15
H3BO3 49.7 mM; NaOH 20.0 mM
9.0
9.18
28.7
14 – 15
H3BO3 29.4 mM; NaOH 20.0 mM
9.5
9.59
15.0
14 – 15
H3BO3 22.8 mM; NaOH 20.0 mM
10.0
9.97
6.2
14 – 15
CAPS 33.1 mM; NaOH 20.0 mM
10.5
10.84
19.8
12 – 13
CAPS 22.9 mM; NaOH 20.0 mM
11.0
11.31
10.6
14 – 15
H3PO4 4.4 mM; NaOH 13.5 mM
11.5
11.81
10.8
17 – 18
H3PO4 1.9 mM; NaOH 16.3 mM
12.0
12.25
29.6
32 – 34
2.2 Sample preparation
Twenty-one buffers were prepared from pH 2.0 to 12.0 with an increment of 0.5 pH unity. They were set at a constant ionic strength of 20 mM, and are listed in Table 1. Their composition was calculated by the Phoebus software 1.0 (Sedere, Centre Analyse, Orleans, France), and the pH value was measured with a Mettler-Toledo SevenMulti pH meter (Schwerzenbach, Switzerland), daily calibrated with three aqueous solutions set at pH 2.00, 4.00, 7.00 from Fluka and one solution at pH 10.00 from Reactolab (Servion, Switzerland).
with a UV detector, an autosampler, and a power supply able to deliver up to 30 kV. The separation was performed in a fused-silica capillary (Polymicro, Phoenix, AZ, USA) with an inner diameter of 50 lm and 67.0 cm total length (60.0 cm to the UV detector). All experiments were carried out in the cationic mode (anode at the inlet and cathode at the outlet). During analysis, a voltage of 30 kV (initial ramping of 3.0 kV/s) was applied. The capillary was thermostated at 258C, samples were kept at ambient temperature in the autosampler, and 1% of the total capillary length was injected by applying a pressure of 0.5 psi (ca. 3.5 kPa) for 15 s. UV detection was carried out at 254 nm. For each pH buffer, the fused-silica capillary was sequentially washed prior analyses with 0.1 M NaOH (3 min), water (3 min), Ceofix initiator (1 min), Ceofix accelerator (2 min), and BGE (5 min). Between analyses, the capillary was flushed with BGE for 3 min. Three runs were performed for each sample at each pH. BGE solutions were replaced after nine injections, to avoid degradation due to oxydoreduction phenomena.
2.4 CE instrumentation
2.5 Procedure
Experiments were performed with a Beckman P/ACE 5000 (Beckman Coulter, Nyon, Switzerland) equipped
The same procedure was applied to the 15 tested compounds. Each drug was injected three times at each pH.
Fifteen samples were prepared. Each sample contained one substance diluted in water/methanol/acetone 94 : 5 : 1 v/v/v. Acetone was used to determine EOF, and methanol was added to ensure better solubility of the tested compounds. Samples were set at different concentrations, from 10 to 1000 g/L, depending on their UV response.
2.3 Buffer preparation
J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Determination of pKa values by CZE
2377
Table 2. Measured pK a values of acidic compounds pK a by pK a by CZE with dynamic potentiometry coating Salicylic acid
2.93 –
pK a from [25]
– –
3.0 13.4
Benzoic acid
4.19
3.98
4.2
Ketoprofen
4.22
4.25 [17]
–
Phenobarbital
7.46
7.20 [19]
7.4
10.05
9.99 [17]
10.0
Phenol
phenobarbital, and phenol. They possess one pK a within the 2.0 – 12.0 pH range, and exhibit similar trends. As an example, Fig. 2A illustrates leff variation of ketoprofen as a function of pH. At pH = 2, ketoprofen was neutral, migrated simultaneously with the EOF marker, and leff was consequently equal to zero. At a pH superior to 6, it was fully deprotonated, and possessed a constant leff value of – 2610 – 4 cm2/V6s. Ketoprofen was half deprotonated at leff equal to – 1610 – 4 cm2/V6s, at pH 4.16 (i. e., pK 9a value). So far, pK 9a and leff values were dependent on CZE conditions. In order to adjust this pK 9a, a correction activity was calculated from Eq. (3) (0.06 and 0.19 unity for monoacids and diacids, respectively) and added to obtain the true pK a of the molecule (Eq. (4)). This correction was carried out for all acidic compounds; pK a values are reported in Table 2. Figure 2. Relationship between leff and pH of BGE for (A) an acidic compound, ketoprofen, and (B) a dibasic compound, quinine.
3.2 Basic compounds
For each injection, leff was calculated from Eq. (6), and reported as a function of pH, giving rise to a sigmoidal curve (Eqs. (4), (5)). With a GraphPad Prism 4.0 software (GraphPad Software, San Diego, CA, USA), a nonlinear regression was performed on 63 points to determine pK 9a values, which is dependent on method conditions (i. e., BGE concentration). The latter values were modified with an activity coefficient so as to obtain the corrected pK a value, independent of the determination procedure.
The four basic compounds selected were lidocaine, propafenone, propranolol, and quinine. The latter possesses two pK a values, and Fig. 2 B illustrates its leff variation with pH. Quinine was neutral at a pH superior to 10 (leff = 0), monoprotonated at pH 6 (leff = 1.7610 – 4 cm2/V6s), and became diprotonated at pH 2 (leff = 3.4610 – 4 cm2/V6 s). Hence, two pK 9a values were determined for quinine. After correction with the activity coefficient (its subtraction, according to Eq. (5)), pK a values, independent of CZE conditions, are reported in Table 3.
3 Results
3.3 Ampholytic and zwitterionic compounds
Results are successively presented for three classes of compounds: five acidic compounds, four basic compounds, and finally one zwitterionic and ampholytic compounds.
The five studied sulfonamides were sulfanilamide, sulfabenzamide, sulfadimethoxine, sulfadoxine, and sulfisoxazole. They are all ampholytes with pK a values superior to 4 and inferior to 2, for their acidic and basic functions, respectively. As exhibited for sulfadoxine (Fig. 3A), the acidic function possesses a pK a value comprised in the investigated pH range, which is not the case for the basic function. However, an estimation of the latter value was carried out.
3.1 Acidic compounds pK a values were determined by CZE for five acidic compounds, namely, salicylic acid, benzoic acid, ketoprofen, J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2378
Geiser, Henchoz, Galland, Garrupt, Veuthey Table 4. Measured pK a values of ampholytic and zwitterionic compounds pK a by CZE pKa by pK a with dynamic potentiometry from [25] coating Sulfanilamide
1.78a) 11.19
– 10.24
– 10.4
Sulfabenzamide
1.20a) 4.36
– 4.20
– –
Sulfadimethoxine
1.62a) 6.13
– 5.96
– 5.9
Sulfadoxine
1.52a) 6.01
– 5.81
– –
Sulfisoxazole
1.32a) 5.00
– 4.79
– 5.0
Cetirizine
2.52b) – 8.21
2.19 2.93 8.00 [16]
– – –
a)
b)
pK a value obtained by extrapolation, at a pH outside the investigated range. Average pK 9a value of acidic and basic functions, uncorrected by the coefficient activity.
basic function), and to 9610 – 5 cm2/V6s at pK 9a value. The latter was estimated at 1.58, giving a pK a of 1.52 after correction with the activity coefficient (Eq. (5)). The same procedure was applied to the four other sulfonamide compounds (Fig. 4); results are reported in Table 4. Figure 3. Relationship between leff and pH of BGE for (A) an ampholyte compound, sulfadoxine, and (B) a zwitterionic compound, cetirizine. Function extrapolation below pH 2.0 is represented with dashed lines. Table 3. Measured pK a values of basic compounds pK a by pK a by CZE with dynamic potentiometry coating
pK a from [25]
Lidocaine
8.09
7.94
7.9
Propafenone
9.48
9.43 [18]
–
Propranolol
9.71
9.53 [19]
9.5
Quinine
3.90 8.41
4.46 8.52 [18]
4.1 8.5
For compounds with two ionizable functions, we observed a Dleff of equal intensity for both pK a values, as exhibited for quinine (Section 3.2). Accordingly, the Dleff intensity of the sulfonamides basic function was estimated as equal to Dleff intensity of the acidic function. As an example, sulfadoxine possesses a leff value of – 1.8610 – 4 cm2/V6s when the acidic function is fully deprotonated at a pH superior to 8 (Fig. 3 A). Therefore, leff was approximately equal to 1.8610 – 4 cm2/V6s at pH = 0 (full protonation of J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
Cetirizine was selected as model zwitterionic compound. It possesses two basic and one acidic functions [16]. The relationship of its mobility with pH is illustrated in Fig. 3 B, exhibiting only two pK 9a values. The Dleff occurring between pH 7 and 9 corresponded to the first protonation of the piperazine function. After correction with the activity coefficient, its pK a value was reported in Table 4. The Dleff which occurred between pH 2 and 4 was two times larger. It corresponded simultaneously to the second protonation of piperazine function and the carboxylic deprotonation. Since the distinction between both pK 9a was ambiguous, the uncorrected pK 9a value was reported in Table 4.
4 Discussion pK a values by CZE are discussed and compared to values determined by other methods. In particular, they are almost always compared with pK a values obtained by potentiometric titrations, according to a procedure described elsewhere (e. g., [16 – 19]).
4.1 pK a determination of acids, bases, and zwitterion For monoacidic and monobasic compounds, pK a values determined by CZE are in accordance with the literature,
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Determination of pKa values by CZE
2379
Figure 4. Relationship between leff of sulfonamides and pH of BGE. As in Fig. 3 A, the function is extrapolated below pH 2.0, for (A) sulfanilamide, (B) sulfabenzamide, (C) sulfadimethoxine, and (D) sulfisoxazole.
as seen in Tables 2 – 4. Confidence intervals, related with the incertitude associated to the pK 9a determination from the curve leff as a function of pH, were around 0.1. However, this confidence interval did not include uncertainties on BGE preparation, including pH measurements. Therefore, it seems more realistic to associate a confidence interval of 0.20 to any pK a value. In particular cases, pK a values could not be determined. This occurred when the pK a value was outside the investigated pH range, and when the molecule possessed two close pK a values. As an example, salicylic acid possesses two acidic functions, one of them with a pK a value superior to 13.0. Since the investigated pH range was between 2.0 and 12.0, it neither could be determined nor estimated (data not shown). Besides, cetirizine was the only studied compound which possessed three pK a within the investigated range. While the pK a of the first basic function (piperazine) was determined at 8.2, the second basic function and the acidic function could not be distinguished. In fact, the difference between both pK a values J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
was probably inferior to 1.0, and only an average pK 9a value (macroscopic pK a) around 2.5 could be assigned. It is worthy to mention that Dleff between pH 2.0 and 3.5 was twice superior to the variation measured between pH 7.5 and 9.0 (Fig. 3B), which clearly indicated two pK a values.
4.2 pK a determination of sulfonamides The five selected sulfonamides are ampholytes with pK a values largely superior for acidic functions than for basic ones. Consequently, no overlap between these two values occurred, and the pK a values of acidic functions were close to those reported in the literature, except for sulfonamide (Table 4). For the latter, uncertainties on the pH measurements at high pH may be an explanation. Further studies will be performed to minimize this drawback. On the other hand, pK a determination of basic functions induced some difficulties [20] because these values were inferior to 2.0. Since a leff increase was observed in Figs.
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2380
Geiser, Henchoz, Galland, Garrupt, Veuthey
3 A, 4 for all sulfonamides, an approximation of pK a could be carried out (Section 3.3). This process was an interesting alternative to the one extending the investigated pH range [21], as the latter leads to theoretical and practical drawbacks. For instance, a BGE of ionic strength superior to 20 mM is necessary for pH values inferior to 2.0. In order to prevent an electric current elevation and Joule effects, a lower voltage must be applied, increasing analysis time. Moreover, the proton activity coefficient has to be taken into account at a concentration higher than 0.01 M, i. e., at a pH inferior to 2.0.
4.3 Analysis time with dynamic coating The dynamic coating procedure used to accelerate the determination of migration time is a generic approach which allows saving time. Migration of acetone (the neutral marker) was carried out in 6 – 8 min, which is very stable considering the large pH range investigated (2.0 – 12.0). Basic compounds were carried out in less than 8 min. More analysis time was necessary for acidic compounds, since the latter migrated after the EOF when deprotonated. It took about 15 min for ketoprofen and phenobarbital, and approximately 30 min for small acidic compounds, such as salicylic acid, benzoic acid, and phenol. For the latter, dynamic coating could be used in combination with other approaches to further reduce analysis time, such as capillary length reduction, short-end injection, or pressure-assisted analyses. In addition to reducing analysis time, the dynamic coating procedure had further advantages. First, Joule heating effects were probably negligible, as illustrated by the generated current value, always inferior to 50 lA (Table 1). Moreover, dynamic coating has been described as (i) improving efficiency due to lEOF increase [22], (ii) stabilizing migration times [23], and (iii) preventing interactions between compounds and capillary walls [24]. Consequently, precision and accuracy on t m and t mn measurements were improved, which led to accurate pK a determination.
4.4 Concluding remarks CZE provided pK a values close to those reported in the literature. Compared to other approaches, the determination of pK a values by CZE presents several advantages, such as small sample consumption as well as the ability to handle compounds containing impurities or which are poorly soluble in water. The use of a dynamic coating procedure afforded further advantages. In fact, lEOF were faster and not as dependent on the BGE pH as without dynamic coating, thus inducing reduced and similar anal-
J. Sep. Sci. 2005, 28, 2374 – 2380
www.jss-journal.de
ysis times. Furthermore, higher efficiencies and better precision on the measured mobilities can be achieved. Consequently, CZE with a dynamic coating procedure can be highly recommended as a generic approach to determine pK a values.
5 References [1] Kibbey, C. E., Poole, S. K., Robinson, B., Jackson, J. D., Durham, D., J. Pharm. Sci. 2001, 90, 1164 – 1175. [2] Poole, S. K., Patel, S., Dehring, K., Workman, H., Poole, C. F., J. Chromatogr. A 2004, 1037, 445 – 454. [3] Beckers, J. L., Everaerts, F. M., Ackermans, M. T., J. Chromatogr. 1991, 537, 407 – 428. [4] Debye, P., Huckel, E., Phys. Z. 1923, 24, 305 – 325. [5] Guntelberg, E., Z. Phys. Chem. 1926, 123, 199 – 247. [6] Cai, J., Smith, J. T., Rassi, Z. E., J. High Resolut. Chromatogr. 1992, 15, 30 – 32. [7] Caliaro, G. A., Herbots, C. A., J. Pharm. Biomed. Anal. 2001, 26, 427 – 434. [8] Cleveland, J. A. J., Benko, M. H., Gluck, S. J., Walbroehl, Y. M., J. Chromatogr. 1993, 652, 301 – 308. [9] Gluck, S. J., Cleveland, J. A. J., J. Chromatogr. A 1994, 680, 43 – 48. [10] Gong, S., Su, X., Bo, T., Zhang, X., Liu, H., Li, K. A., J. Sep. Sci. 2003, 26, 549 – 554. [11] Gluck, S. J., Steele, K. P., Benkoe, M. H., J. Chromatogr. A 1996, 745, 117 – 125. [12] Wan, H., Holmen, A., Nagard, M., Lindberg, W., J. Chromatogr. A 2002, 979, 369 – 377. [13] Ishihama, Y., Nakamura, M., Miwa, T., Kajima, T., Asakawa, N., J. Pharm. Sci. 2002, 91, 933 – 942. [14] Jia, Z., Ramstad, T., Zhong, M., Electrophoresis 2001, 22, 1112 – 1118. [15] Miller, J. M., Blackburn, A. C., Shi, Y., Melzak, A. J., Ando, H. Y., Electrophoresis 2002, 23, 2833 – 2841. [16] Bouchard, G., Pagliara, A., Van Balen, G. P., Carrupt, P. A., Testa, B., Gobry, V., Girault, H. H. et al., Helv. Chim. Acta 2001, 84, 375 – 387. [17] Bouchard, G., Carrupt, P. A., Testa, B., Gobry, V., Girault, H. H., Chem.-Eur. J. 2002, 8, 3478 – 3484. [18] Plemper van Balen, G., Carrupt, P. A., Morin, D., Tillement, J. P., Le Ridant, A., Testa, B., Biochem. Pharmacol. 2002, 63, 1691 – 1697. [19] Taillardat-Bertschinger, A., Martinet, C. A. M., Carrupt, P. A., Reist, M., Caron, G., Fruttero, R., Testa, B., Pharm. Res. 2002, 19, 729 – 737. [20] Cross, R. F., Ricci, M. C., LC-GC 1995, 13, 132 – 142. [21] Mercier, J. P., Morin, P., Dreux, M., Tambute, A., Chromatographia 1998, 48, 529 – 534. [22] Kenndler, E., J. Capil. Electrophor. 1996, 3, 191 – 198. [23] Wehr, T., LC-GC North America 2001, 19, 398 – 407. [24] Clohs, L., Winstanley, A. K., LC-GC Europe 2002, 15, 288 – 296. [25] Moffat, A. C., Jackson, J. V., Moss, M. S., Widdop, B., Greenfield, E. S., Clarke’s Isolation and Identification of Drugs, 2nd Edn., The Pharmaceutical Press, London 1986.
i
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
